講笑話 - a context sensitive phrase

[henrymcl]

Describe the bug

講笑話 can be parsed as 講笑/話 and 講/笑話

>>> pycantonese.characters_to_jyutping('笑話')
[('笑話', 'siu3waa2')]
>>> pycantonese.characters_to_jyutping('佢好鐘意講笑話佢女朋友肥')
[('佢', 'keoi5'), ('好', 'hou2'), ('鐘意', 'zung1ji3'), ('講笑', 'gong2siu3'), ('話', 'waa6'), ('佢', 'keoi5'), ('女朋友', 'neoi2pang4jau5'), ('肥', 'fei4')]
>>> pycantonese.characters_to_jyutping('佢好鐘意講笑話')
[('佢', 'keoi5'), ('好', 'hou2'), ('鐘意', 'zung1ji3'), ('講笑', 'gong2siu3'), ('話', 'waa6')]

Expected behavior

>>> pycantonese.characters_to_jyutping('佢好鐘意講笑話')
[('佢', 'keoi5'), ('好', 'hou2'), ('鐘意', 'zung1ji3'), ('講', 'gong2'), ('笑話', 'siu3waa2')]

There isn't probably that many other similar cases (perhaps even none) as it involves the quoting word 話 interacting with 講笑 in multiple ways.

Screenshots
If applicable, add screenshots to help explain your problem.

System (please complete the following information):

System: Ubuntu 22.04.5 LTS

Output from pip list:

certifi            2025.10.5
charset-normalizer 3.4.3
idna               3.10
pip                22.0.2
pycantonese        3.4.0
pylangacq          0.16.2
python-dateutil    2.9.0.post0
requests           2.32.5
setuptools         59.6.0
six                1.17.0
tabulate           0.9.0
urllib3            2.5.0
wcwidth            0.2.14
wordseg            0.0.2

Additional context
Add any other context about the problem here.